Techniques facilitating detection of conversation threads in unstructured channels are provided. A system can comprise a memory that stores computer executable components and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise an extraction component that employs a model to detect conversation messages based on a defined confidence level and assigns the conversation messages to respective conversation thread categories. The computer executable components also can comprise a model component that trains the model on conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to the defined confidence level.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system, comprising: a memory that stores computer executable components; and a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise: a model component that trains, using machine learning, a model on a first set of unstructured conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to a defined confidence level; and an extraction component that employs the model to transform a second set of unstructured conversation messages from a plurality of parties communicating in a communication channel into structured conversation threads having respective contexts, wherein the model: for a first subset of first unstructured conversation messages of the second set of unstructured conversation messages performs a pairwise sentence comparison of the first unstructured conversation messages to assign the first unstructured conversation messages to the structured conversation threads, and for a second subset of second unstructured conversation messages of the second set of unstructured conversation messages: performs a context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, and based on the context similarity comparison, for respective second unstructured conversation messages, at least one of assign the the second unstructured conversation message to one of the structured conversation threads or assign the second unstructured conversation message to a new structured conversation thread, wherein the context similarity comparison does not comprise the pairwise sentence comparison involving the second unstructured conversation messages, and the first subset is different than the second subset.
2. The system of claim 1 , wherein the context similarity comparison generates respective context scores between the second unstructured conversation messages and the respective contexts.
The system relates to analyzing unstructured conversation messages to determine their relevance to specific contexts. The problem addressed is the difficulty in accurately matching unstructured text, such as chat messages or emails, to predefined contexts or topics, which is crucial for applications like customer support, content moderation, or automated response systems. The system includes a processing module that receives unstructured conversation messages and compares them to predefined contexts. A context similarity comparison module generates context scores for each message, indicating how closely the message aligns with each context. These scores help prioritize or categorize messages based on their relevance to different contexts, improving efficiency in handling large volumes of unstructured data. The system may also include a message selection module that filters or ranks messages based on their context scores, ensuring that the most relevant messages are processed first. Additionally, a context database stores predefined contexts, which may be updated dynamically to reflect evolving topics or user needs. The system may further include a user interface for displaying context scores and allowing users to refine or adjust the context matching process. By quantifying the relevance of messages to specific contexts, the system enhances the accuracy and efficiency of automated text analysis, reducing manual effort and improving decision-making in applications that rely on unstructured data.
3. The system of claim 1 , wherein the model component trains the model to identify the respective text data of the first set of unstructured conversation messages received over a defined interval.
The system relates to natural language processing and machine learning for analyzing unstructured conversation data. The problem addressed is the difficulty in extracting meaningful insights from large volumes of unstructured text-based conversations, such as chat logs or social media exchanges, due to their lack of predefined structure. The system includes a model component that trains a machine learning model to identify and categorize text data from a first set of unstructured conversation messages. The training process focuses on messages received over a defined time interval, allowing the model to adapt to evolving language patterns or topics. The model component may use techniques such as natural language processing, deep learning, or statistical analysis to process the text data. The system may also include a preprocessing module to clean or normalize the text data before training, and an output module to generate structured insights or summaries from the analyzed conversations. The trained model can then be applied to new, incoming messages to classify or extract relevant information automatically. This approach enables real-time or batch analysis of conversation data for applications like customer support, sentiment analysis, or trend detection.
4. The system of claim 1 , wherein the model component trains the model on the respective text data of the first set of unstructured conversation messages that commenced during a defined interval.
This invention relates to a system for training machine learning models on unstructured conversation data, specifically focusing on messages exchanged within a defined time interval. The system addresses the challenge of efficiently processing and analyzing large volumes of unstructured conversational data to improve model accuracy and relevance. The core system includes a model component that processes text data from a first set of unstructured conversation messages, where these messages are selected based on their initiation time falling within a specified time window. The model component trains a machine learning model using this filtered dataset, enabling the system to capture temporal patterns and contextual relevance within the conversations. The system may also include a data processing component that preprocesses the conversation messages, such as by cleaning, normalizing, or extracting features from the text data. Additionally, the system may incorporate a user interface component that allows users to define the time interval for message selection, adjust training parameters, or visualize training results. The invention aims to enhance the performance of conversational AI models by leveraging time-bound data segments, improving the model's ability to understand and respond to dynamic conversational contexts.
5. The system of claim 4 , wherein the model component ignores prior unstructured conversation messages that commenced prior to the defined interval for training the model.
The invention relates to a system for training a machine learning model using conversation data, addressing the challenge of efficiently processing large volumes of unstructured conversational messages while ensuring the model is trained on relevant and timely data. The system includes a model component that filters out prior unstructured conversation messages that commenced before a defined interval, ensuring that only recent or relevant conversations are used for training. This filtering mechanism helps improve the model's performance by excluding outdated or irrelevant data, which could otherwise introduce noise or bias. The system may also include a data processing component that extracts and preprocesses features from the conversation messages, and a training component that uses the filtered data to train the model. The defined interval can be dynamically adjusted based on factors such as conversation relevance, user feedback, or system performance metrics. By focusing on recent or contextually relevant conversations, the system enhances the accuracy and efficiency of the trained model in applications such as chatbots, virtual assistants, or customer support systems.
6. The system of claim 1 , wherein the second set of unstructured conversation messages comprise parallel conversations that occur during an overlapping time period.
The invention relates to a system for analyzing unstructured conversation messages, particularly in scenarios where multiple parallel conversations occur simultaneously or during overlapping time periods. The system is designed to process and extract meaningful insights from these conversations, which may otherwise be difficult to analyze due to their unstructured nature and the complexity introduced by concurrent discussions. The system includes a processing module that receives and categorizes conversation messages into at least two sets. The first set consists of messages from a primary conversation, while the second set includes messages from parallel conversations that occur during the same or overlapping time periods as the primary conversation. The system is capable of distinguishing between these sets and analyzing them independently or in relation to each other, depending on the context. By identifying and separating parallel conversations, the system can reduce noise and improve the accuracy of insights derived from the primary conversation. This is particularly useful in environments such as customer service, team collaboration, or social media monitoring, where multiple discussions may intersect or influence each other. The system may also include additional features, such as natural language processing (NLP) or machine learning algorithms, to further enhance the analysis of these conversations.
7. The system of claim 1 , wherein the pairwise sentence comparison generates a sentence similarity score.
The system relates to natural language processing (NLP) and text analysis, specifically addressing the challenge of comparing sentences to determine their semantic similarity. The system includes a method for analyzing text by breaking it into sentences, then comparing each sentence to every other sentence in the text to generate a similarity score. This pairwise comparison helps identify relationships between sentences, such as redundancy, paraphrasing, or thematic connections. The system may also include preprocessing steps like tokenization, normalization, or embedding generation to prepare sentences for comparison. The similarity score is computed using techniques like cosine similarity, word embeddings, or neural network models, allowing the system to quantify how closely related two sentences are. This enables applications in text summarization, plagiarism detection, or content analysis by identifying semantically similar sentences within a document. The system may further include a threshold mechanism to filter out low-similarity pairs, improving efficiency and relevance. The overall goal is to provide an automated way to assess sentence-level similarity in large text corpora, aiding in tasks like document clustering, information retrieval, or content recommendation.
8. The system of claim 1 , wherein the performing the context similarity comparison comprises: performing the context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, comprising: in response to the second unstructured conversation message matching, according to a similarity criterion, a context of the respective contexts, assigning the second unstructured conversation message to a first structured conversation thread associated with the context, and in response to the second unstructured conversation message not matching, according to the similarity criterion, the respective contexts, assigning the second unstructured conversation message to the new structured conversation thread associated with a new context.
The invention relates to a system for organizing unstructured conversation messages into structured conversation threads based on context similarity. The problem addressed is the difficulty of automatically categorizing and grouping unstructured messages, such as those in chat applications or forums, into meaningful, contextually relevant threads without manual intervention. The system processes unstructured conversation messages by comparing their content to the contexts of existing structured conversation threads. When a message matches the context of an existing thread according to a predefined similarity criterion, it is assigned to that thread. If no match is found, the message is used to create a new structured conversation thread with a new context. The similarity criterion may involve natural language processing, keyword matching, or other techniques to determine contextual relevance. This approach ensures that related messages are grouped together, improving organization and readability in digital communication platforms. The system dynamically adapts to new topics by creating new threads when necessary, while maintaining coherence within existing discussions. The method is particularly useful in environments where large volumes of unstructured messages need to be systematically organized for better user experience or data analysis.
9. A computer-implemented method, comprising: training, by a system operatively coupled to a processor, using machine learning, a model on a first set of unstructured conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to a defined confidence level; employing, by the system, the model on a first subset of first unstructured conversation messages of a second set of unstructured conversation messages from a plurality of parties communicating in a communication channel to perform a pairwise sentence comparison of the first unstructured conversation messages to assign the first unstructured conversation messages to structured conversation threads; and employing, by the system, the model on a second subset of second unstructured conversation messages of the second set of unstructured conversation messages to: perform a context similarity comparison of the second unstructured conversation messages to respective contexts of the structured conversation threads, and based on the context similarity comparison, for respective second unstructured conversation messages, at least one of assign the second unstructured conversation message to one of the structured conversation threads or assign the second unstructured conversation message to a new structured conversation thread, wherein the context similarity comparison does not comprise the pairwise sentence comparison involving the second unstructured conversation messages, and the first subset is different than the second subset.
This invention relates to a machine learning-based system for organizing unstructured conversation messages into structured conversation threads. The problem addressed is the difficulty of automatically categorizing and threading unstructured text data from multi-party communications, such as chat logs or messaging platforms, into coherent and contextually relevant threads. The system trains a machine learning model on a first set of unstructured conversation messages to detect text data with a defined confidence level. The trained model is then applied to a second set of unstructured messages from a communication channel involving multiple parties. The model first processes a subset of these messages using pairwise sentence comparison to assign them to structured conversation threads. A different subset of messages is then processed using a context similarity comparison to either assign them to existing threads or create new threads, based on their relevance to the context of the structured threads. The context similarity comparison does not involve pairwise sentence comparison for these messages, ensuring efficient and accurate threading. The system improves the organization of multi-party conversations by dynamically structuring messages into meaningful threads while distinguishing between different types of message processing for optimal performance.
10. The computer-implemented method of claim 9 , wherein the context similarity comparison generates respective context scores between the second unstructured conversation messages and the respective contexts.
This invention relates to analyzing unstructured conversation messages, such as those in chat logs or customer support interactions, to determine their relevance to specific contexts. The problem addressed is the difficulty of automatically categorizing or routing unstructured text data based on contextual relevance, which is crucial for applications like customer service automation, sentiment analysis, or knowledge management. The method involves processing a set of unstructured conversation messages to extract features, such as keywords, phrases, or semantic relationships. These features are then compared to predefined contexts, which may represent topics, intents, or categories of interest. A context similarity comparison is performed to generate context scores, which quantify how closely each message aligns with each context. These scores help determine the most relevant context for each message, enabling automated classification, prioritization, or routing of the conversations. The method may also include preprocessing steps, such as normalizing text, removing noise, or applying natural language processing (NLP) techniques to improve feature extraction. The context similarity comparison may use techniques like cosine similarity, machine learning models, or rule-based matching to compute the scores. The resulting context scores can be used to filter, rank, or group messages based on their relevance to different contexts, improving efficiency in handling large volumes of unstructured data.
11. The computer-implemented method of claim 9 , wherein the training the model further comprises training, by the system using the machine learning, the model to identify the respective text data of the first set of unstructured conversation messages received over a defined interval.
This invention relates to a computer-implemented method for training a machine learning model to analyze unstructured conversation data. The method addresses the challenge of extracting meaningful insights from large volumes of unstructured text, such as chat logs or messaging data, by enabling a system to identify and process relevant text segments within a defined time interval. The system first receives a set of unstructured conversation messages, which may include text from various sources like customer support chats, social media interactions, or internal team communications. These messages are processed to extract features that help the model understand the context and content of the conversations. The model is then trained using machine learning techniques to recognize and categorize specific text data within these messages, particularly focusing on segments received over a specified time period. This allows the system to detect patterns, trends, or key information that may be time-sensitive or relevant to a particular analysis. The training process involves adjusting the model's parameters based on feedback or labeled data to improve its accuracy in identifying the desired text segments. The system may also incorporate additional data, such as metadata or user interactions, to enhance the model's ability to distinguish between different types of messages or conversations. By focusing on a defined interval, the method ensures that the model can adapt to real-time or time-bound analysis requirements, making it useful for applications like sentiment analysis, customer feedback monitoring, or automated response generation. The overall approach improves the efficiency and accuracy of extracting actionable insights from unstructured conversational data.
12. The computer-implemented method of claim 9 , wherein the training the model further comprises training, by the system, the model on the respective text data of the first set of unstructured conversation messages that commenced during a defined interval.
This invention relates to a computer-implemented method for training a machine learning model on unstructured conversation data. The method addresses the challenge of improving model performance by leveraging temporal patterns in conversational data. Specifically, the system trains a model on a first set of unstructured conversation messages that were initiated within a defined time interval. This approach allows the model to capture contextual and temporal dependencies that may exist within conversations that start close together, enhancing its ability to understand and generate relevant responses. The training process involves processing the text data of these conversations to extract meaningful features and patterns, which are then used to refine the model's predictive capabilities. By focusing on conversations that commenced during a specific time window, the system can better adapt to dynamic changes in language use, user behavior, or topic trends. This method is particularly useful in applications such as chatbots, virtual assistants, and customer support systems, where understanding the context and timing of conversations is critical for providing accurate and timely responses. The invention improves upon existing techniques by incorporating temporal constraints in the training data, leading to more contextually aware and responsive models.
13. The computer-implemented method of claim 12 , further comprising: ignoring, by the system, prior unstructured conversation messages that commenced prior to the defined interval for training the model.
This invention relates to a computer-implemented method for training a conversational model using structured and unstructured conversation data. The method addresses the challenge of improving model accuracy by selectively filtering training data to exclude irrelevant or outdated information. The system defines a time interval for training the model, then processes both structured and unstructured conversation messages within that interval. Structured messages are those with predefined formats, such as forms or templates, while unstructured messages lack such formatting. The system extracts features from these messages, including metadata like timestamps, participant identifiers, and message content, and uses these features to train the model. Additionally, the system ignores prior unstructured conversation messages that began before the defined interval, ensuring the model is trained only on recent and relevant data. This selective filtering helps improve the model's performance by reducing noise from outdated or irrelevant conversations. The method may also involve preprocessing the messages, such as normalizing text or removing sensitive information, before feature extraction. The trained model can then be deployed to analyze or generate responses in new conversations, leveraging the filtered and structured data for better accuracy.
14. The computer-implemented method of claim 9 , wherein the second set of unstructured conversation messages comprise parallel conversations that occur during an overlapping time period.
This invention relates to analyzing unstructured conversation messages, particularly in scenarios where multiple parallel conversations occur simultaneously. The method involves processing a first set of unstructured conversation messages to identify a first topic, then analyzing a second set of unstructured conversation messages to identify a second topic. The second set of messages represents parallel conversations that occur during an overlapping time period with the first set. The method further includes determining a relationship between the first and second topics, such as whether they are related or unrelated, and generating an output based on this relationship. This output may include a summary, a recommendation, or an actionable insight derived from the analysis. The technique is useful in environments where multiple discussions happen concurrently, such as in customer support systems, social media monitoring, or collaborative workspaces, where understanding the interplay between parallel conversations can improve decision-making or automation. The method may also involve filtering or prioritizing conversations based on their relevance to the identified topics.
15. The computer-implemented method of claim 9 , wherein the pairwise sentence comparison generates a sentence similarity score.
This invention relates to natural language processing (NLP) and text analysis, specifically improving the accuracy of comparing sentences to determine their similarity. The problem addressed is the need for more precise and efficient methods to assess how similar two sentences are, which is critical for applications like document summarization, plagiarism detection, and information retrieval. Existing methods often struggle with contextual nuances, synonyms, or structural differences that can obscure true semantic similarity. The method involves a pairwise sentence comparison process that generates a sentence similarity score, quantifying how closely related two sentences are. This comparison likely involves analyzing linguistic features, semantic relationships, or machine learning models trained to recognize semantic equivalence despite variations in wording. The score can be used to cluster similar sentences, identify duplicates, or rank documents by relevance. The technique may also incorporate preprocessing steps like tokenization, part-of-speech tagging, or embedding generation to enhance comparison accuracy. By producing a numerical similarity score, the method provides a measurable way to assess sentence relationships, improving tasks that rely on text similarity analysis. The approach is particularly useful in automated systems where manual review is impractical, ensuring consistency and scalability in text processing workflows.
16. The computer-implemented method of claim 9 , wherein the performing the context similarity comparison comprises: performing, by the system using the model, the context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, comprising: in response to the second unstructured conversation message matching, according to a similarity criterion, a context of the respective contexts, assigning the second unstructured conversation message to a first structured conversation thread associated with the context, and in response to the second unstructured conversation message not matching, according to the similarity criterion, the respective contexts, assigning the second unstructured conversation message to the new structured conversation thread associated with a new context.
This invention relates to a computer-implemented method for organizing unstructured conversation messages into structured conversation threads based on context similarity. The problem addressed is the difficulty of automatically categorizing and grouping unstructured messages, such as those in chat systems or forums, into meaningful, contextually relevant threads without manual intervention. The method involves using a trained model to compare the context of unstructured messages against existing structured conversation threads. When an unstructured message matches the context of an existing thread according to a predefined similarity criterion, it is assigned to that thread. If no match is found, the message is used to create a new structured thread with a new context. The system dynamically updates the structured threads as new messages are processed, ensuring that conversations remain organized and relevant. The model evaluates the semantic and contextual relevance of messages, allowing for accurate grouping even when messages lack explicit structural indicators. This approach improves information retrieval, user experience, and automated moderation in digital communication platforms. The method is particularly useful in environments where large volumes of unstructured data must be efficiently organized for analysis or user interaction.
17. A computer program product that facilitates detection of conversation threads in a messaging channel, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions are executable by a processor to cause the processor to: training, using machine learning, a model on a first set of unstructured conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to a defined confidence level; employ the model on a first subset of first unstructured conversation messages of a second set of unstructured conversation messages from a plurality of parties communicating in a communication channel to perform a pairwise sentence comparison of the first unstructured conversation messages to assign the first unstructured conversation messages to structured conversation threads; and employ the model on a second subset of second unstructured conversation messages of the second set of unstructured conversation messages to: perform a context similarity comparison of the second unstructured conversation messages to respective contexts of the structured conversation threads, and based on the context similarity comparison, for respective second unstructured conversation messages, at least one of assign the second unstructured conversation message to one of the structured conversation threads or assign the second unstructured conversation message to a new structured conversation thread, wherein the context similarity comparison does not comprise the pairwise sentence comparison involving the second unstructured conversation messages, and the first subset is different than the second subset.
This invention relates to a system for detecting and organizing conversation threads in messaging channels using machine learning. The problem addressed is the difficulty of automatically structuring unstructured messages from multiple participants into coherent conversation threads, which is particularly challenging in group communication channels where messages may lack clear indicators of thread affiliation. The system involves a computer program product that trains a machine learning model on a dataset of unstructured conversation messages to detect and categorize text data with a defined confidence level. The trained model is then applied to a new set of unstructured messages from a communication channel. The process involves two main steps: first, a pairwise sentence comparison is performed on a subset of messages to assign them to structured conversation threads. Second, a different subset of messages undergoes a context similarity comparison to either assign them to existing threads or create new ones. The context similarity comparison does not involve pairwise sentence comparisons, ensuring efficiency and scalability. The system dynamically organizes messages into threads based on content and context, improving readability and organization in group communication channels.
18. The computer program product of claim 17 , wherein the context similarity comparison generates respective context scores between the second unstructured conversation messages and the respective contexts.
This invention relates to natural language processing and conversational systems, specifically addressing the challenge of accurately interpreting and responding to unstructured conversation messages by leveraging contextual similarity. The system analyzes unstructured conversation messages, such as those in chatbots or customer service interactions, to determine their relevance to predefined contexts. A context similarity comparison module generates context scores for each message, quantifying how closely the message aligns with each context. These scores help prioritize or filter messages based on their contextual relevance, improving response accuracy and efficiency. The system may also include preprocessing steps to normalize or extract features from the messages before comparison. The context similarity comparison may use techniques like semantic analysis, keyword matching, or machine learning models to assess relevance. The generated context scores can be used to route messages to appropriate handlers, trigger specific responses, or enhance the overall conversational flow. This approach ensures that unstructured messages are processed in a structured manner, reducing ambiguity and improving system performance in dynamic conversational environments.
19. The computer program product of claim 17 , wherein the pairwise sentence comparison generates a sentence similarity score.
This invention relates to natural language processing (NLP) and text analysis, specifically improving the accuracy of sentence similarity comparisons in computational systems. The problem addressed is the difficulty in accurately determining semantic similarity between sentences, which is crucial for applications like document summarization, machine translation, and information retrieval. Existing methods often struggle with contextual nuances, synonyms, and structural variations, leading to imprecise results. The invention describes a computer program product that enhances sentence similarity analysis by generating a sentence similarity score through pairwise sentence comparisons. This involves processing input text to extract sentences, then comparing each pair of sentences to compute a similarity score based on semantic and syntactic features. The system may use machine learning models, such as neural networks, to analyze word embeddings, syntactic structures, or other linguistic features. The pairwise comparison ensures that every possible sentence pair is evaluated, improving the robustness of the similarity assessment. The resulting score quantifies how closely related two sentences are, enabling more accurate applications in text processing tasks. The invention may also include preprocessing steps like tokenization, normalization, or noise reduction to refine the input before comparison. The output similarity score can be used to cluster similar sentences, identify redundant information, or improve text generation tasks. This approach enhances the precision of NLP applications by providing a reliable metric for sentence similarity.
20. The computer program product of claim 17 , wherein the performance of the context similarity comparison comprises: perform the context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, comprising: in response to the second unstructured conversation message matching, according to a similarity criterion, a context of the respective contexts, assign the second unstructured conversation message to a first structured conversation thread associated with the context, and in response to the second unstructured conversation message not matching, according to the similarity criterion, the respective contexts, assign the second unstructured conversation message to the new structured conversation thread associated with a new context.
This invention relates to a system for organizing unstructured conversation messages into structured conversation threads based on context similarity. The problem addressed is the difficulty of automatically categorizing and grouping unstructured messages, such as those in chat applications or forums, into meaningful, structured threads without manual intervention. The system performs a context similarity comparison between unstructured messages and existing structured conversation threads. Each structured thread has an associated context, which represents the topic or subject of the thread. When a new unstructured message is received, the system compares its content to the contexts of existing threads. If the message matches a context according to a predefined similarity criterion, it is assigned to the corresponding structured thread. If no match is found, the message is assigned to a new structured thread, which is then associated with a new context derived from the message. The similarity criterion may involve natural language processing techniques, such as semantic analysis or keyword matching, to determine how closely the message aligns with the context of existing threads. This automated approach ensures that conversations are logically grouped, improving organization and accessibility in digital communication platforms.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
May 6, 2019
March 1, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.